@john so far I'm *kind of* managing to keep the situation under control on my instance with a more aggressive `tootctl <media|preview_cards> remove` policy - running the cronjob more often, and removing anything older than 3 days.

But even with this policy I've got a 10 GB database dump, and the live-web Docker image is around 7 GB - meaning mostly filled with media stuff posted/viewed just in the last 3 days (!!)

And I've also been careful in associating only to relay servers that don't push a lot of traffic from a lot of instances, otherwise the situation would be more dramatic.

I think it's time that admins start to be a bit less shy about asking users to donate to keep the lights on. #ElonMusk was planning to turn #Twitter profitable, but so far the only one who is likely to see a lot more profit is probably #Linode, thanks to all the nodes being upgraded in these days.